A brain-wide map of descending inputs onto spinal V1 interneurons

Abstract

Motor output results from the coordinated activity of neural circuits distributed across multiple brain regions that convey information to the spinal cord via descending motor pathways. Yet the organizational logic through which supraspinal systems target discrete components of spinal motor circuits remains unclear. Here, using viral transsynaptic tracing along with serial two-photon tomography, we have generated a whole-brain map of monosynaptic inputs to spinal V1 interneurons, a major inhibitory population involved in motor control. We identified 26 distinct brain structures that directly innervate V1 interneurons, spanning medullary and pontine regions in the hindbrain as well as cortical, midbrain, cerebellar, and neuromodulatory systems. Moreover, we identified broad but biased input from supraspinal systems onto V1^Foxp2 and V1^Pou6f2 neuronal subsets. Collectively, these studies reveal elements of biased connectivity and convergence in descending inputs to molecularly distinct interneuron subsets and provide an anatomical foundation for understanding how supraspinal systems influence spinal motor circuits.

Keywords: connectivity; interneurons; monosynaptic rabies virus; motor system; neural circuits; spinal cord; supraspinal pathways; viral tracing; whole-brain map.

Figure 1.. Strategy for transsynaptic tracing of descending inputs onto spinal V1 interneurons.

(A) Schematic detailing the experimental design and viral reagents used for brain-wide mapping of inputs onto cervical V1 interneurons. (B) Example of AAV helper virus transduction of V1 interneurons (LacZ, green) in En1::Cre; Rosa26.lsl.LacZ mice infected with AAV-FLEX-TVA-mCherry (TVA, red) and AAV-FLEX-H2B-HA-N2cG (Glycoprotein, blue) at P14 and analyzed at P28. Scale bars = 200 μm (top) or 20 μm (bottom). (C) Specificity of TVA and G expression within V1 interneurons, defined as the percentage of neurons with TVA-mCherry or H2B-HA-N2cG expression that are LacZ⁺ (mean ± SEM; n = 3 animals). (D) Efficiency of viral transduction of V1 interneurons, defined as the percentage of LacZ⁺ neurons that express either TVA, G, or both TVA and G (mean ± SEM; n = 3 animals). (E) Left: Schematic of imaging platforms used for surveying brain-wide inputs onto V1 interneurons (top) and identifying starter cells in the spinal cord (bottom). Middle-left: Registration of brain images to the Allen CCFv3 and segmentation to identify rabies virus (RV)-infected neurons (top), and identification of starter cells in the spinal cord (bottom). Scale bars = 2 mm or 100 μm. Middle-Right: Brain-wide analysis of neurons innervating spinal V1 interneurons. Right: Region-specific analysis of RV-infected neurons. See also Figures S1–S3.

Figure 2.. Identification of brain-wide inputs to spinal V1 interneurons.

(A) Three-dimensional whole-brain visualization of rabies virus (RV)-infected supraspinal neurons providing input to cervical V1 interneurons, registered to the Allen CCFv3 (n = 5 mice; 30,421 neurons). (B) Fraction of RV-infected supraspinal neurons across major brain regions (mean ± SEM, n = 5 mice). (C) Relation between the number of starter cells and that of supraspinal neurons in En1::Cre mice (dotted lines indicate the 95% confidence interval of linear regression). (D) Spatial correlation analysis (cosine distance) of the positions of supraspinal neurons across animals. (E) Fraction of RV-infected supraspinal neurons within each of the 26 identified brain regions, according to the Allen CCFv3 nomenclature (mean ± SEM; n = 5 mice). See also Table S1 for corresponding Paxinos nomenclature, and Figures S2 and S3.

Figure 3.. Anterograde tracing and synaptic physiology demonstrate monosynaptic input from brainstem nuclei to V1 interneurons.

(A) Schematic detailing the strategy for Synaptophysin-eGFP (Syn-eGFP) anterograde tracing from GRN or the vestibular nuclear complex (VNC) onto cervical V1 interneurons. (B) Example of Syn-eGFP inputs arising from Chx10⁺ GRN neurons onto V1 interneurons. Scale bars = 10 μm (left) or 2 μm (right). (C) Quantification of ipsilateral V1 interneurons receiving Syn-eGFP input from the GRN (29.3% ± 2.5%, mean ± SEM, n = 3 animals). (D) Example of Syn-eGFP inputs arising from VNC neurons onto V1 interneurons, as in (B). (E) Quantification of ipsilateral V1 interneurons receiving Syn-eGFP input from the VNC (29.5% ± 4.0%, mean ± SEM, n = 3 animals). (F) Schematic detailing the experimental design used for ChR2-assisted circuit mapping (CRACM) of inputs from Chx10⁺ neurons in the GRN onto cervical V1 interneurons in adult Chx10::Cre; En1::Flpo; RC.fsf.tdT mice. (G) Left, transverse cervical spinal cord from an adult (11 week old) Chx10::Cre; En1::Flpo; RC.fsf.tdT mouse unilaterally injected with AAV-DIO-ChR2-eYFP into the right GRN three weeks earlier. Top right, V1 interneuron with putative GRN synaptic contacts. Bottom right, closeup of synaptic contacts, indicated by arrows. Scale bars = 200 μm (left), 10 μm (top right), or 2 μm (bottom right). (H) Traces of light-evoked excitatory postsynaptic currents (EPSCs) in a V1 interneuron following 2 ms light pulses on GRN axon terminals, across five stimulation trials (1–5). (I) Latency of onset of synaptic responses in V1 interneurons (3.8 ms ± 0.2 ms, mean ± SEM, n = 4 neurons, 3 mice). (J) Jitter of synaptic responses in V1 interneurons (0.24 ms ± 0.02 ms, mean ± SEM, n = 4 neurons, 3 mice). (K) Coefficient of variation of onset (CV_onset) of synaptic responses in V1 interneurons (0.06 ± 0.01, mean ± SEM, n = 4 neurons, 3 mice). See also Figures S4 and S5.

Figure 4.. Hemispheric and region-specific spatial distributions of supraspinal neurons innervating spinal V1 interneurons.

(A) Horizontal view of select brain regions highlighted in (C-H). (B) Normalized cell densities of supraspinal neurons for each brain region, segregated by ipsilateral and contralateral inputs (mean ± SEM; n = 5 mice). (C) Topographic distribution of supraspinal neurons (green) and their corresponding contour plots (black) within sensorimotor cortical regions (MOp, MOs, and SSp) shown in the horizontal plane, with line density curves along the rostrocaudal and mediolateral axes (gray, individual mice; black, mean density). Coordinates represent distance (in mm) from the Allen CCFv3 origin (RC and DV axes) or midline (ML axis). (D-H) Topographic distributions for (D) RN, (E) IP, (F) vestibular nuclei (MV, LAV, SPIV), (G) coronal section of hindbrain at the level of GRN, and (H) GRN, as in panel (C).

Figure 5.. Identification of brain-wide inputs to molecularly defined V1 subsets.

(A) Schematic detailing the experimental design and viral reagents used for brain-wide mapping of inputs onto cervical V1^Foxp2 and V1^Pou6f2 clades. (B) Relation between the number of starter cells and that of supraspinal neurons in En1::Cre; Foxp2::Flpo mice (dotted lines indicate the 95% confidence interval of linear regression). (C) Spatial correlation analysis of soma position for supraspinal neurons innervating V1^Foxp2 interneurons (n = 3 mice). (D) Relation between the number of starter cells and that of supraspinal neurons in En1::Cre; Pou6f2::Flpo mice, as in (B). (E) Spatial correlation analysis of soma positions for supraspinal neurons innervating V1^Pou6f2 interneurons (n = 3 mice). (F) Percentage of RV-infected supraspinal neurons innervating V1^Foxp2 interneurons (dark gray) and V1^Pou6f2 interneurons (blue) versus the general V1 population (light gray, data from Figure 2) within each brain region (mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA followed by Tukey HSD test; n = 3 mice). (G-H) Maximum-intensity STPT projection (750 μm in z-plane) of caudal brainstem showing RV-infected GRN or VNC neurons providing input to V1^Foxp2 (G) or V1^Pou6f2 (H) interneurons. Scale bar = 500 μm (top) or 100 μm (bottom). (I) Example of V1^Foxp2 interneuron with synaptic input (Syn-eGFP, green) from VNC. Scale bar = 10 μm. (J) Example of V1^Pou6f2 interneuron with synaptic input from GRN. Scale bar = 10 μm. See also Figures S3 and S6.